Seizure prediction
Seizure prediction refers to the forecasting of epileptic seizures based on physiological parameters, particularly of the electroencephalogram of epilepsy patients. Background Epileptic seizures are events during which abnormal synchronization of neurons in the brain leads to manifest disturbances in perception, thinking, and behavior. About 0.8% of the world's population suffer from epilepsy which is characterized by recurrent epileptic seizures which occur without obvious external precipitant. So far, the mechanisms leading to the apparently “spontaneous” occurrence of seizures are incompletely understood. Genetic make-up in determining the individual “seizure threshold”, intrinsic fluctuations in EEG dynamics, and external triggers are thought to play a role in the transition from the interictal state to a seizure. The frequency of epileptic seizures may vary from many per day, to a few within several years. Even if epileptic seizures are rare in a given patient, their unforeseeable occurrence overshadows the whole life. The sudden occurrence of seizures endangers the patients during many activities of daily life, often resulting in traumatic injuries and even death e.g. from drowning. Furthermore, being suddenly overwhelmed by loss of control regarding thoughts and behavior may result in serious psychological problems like anxiety, avoidance behaviour and depression, and has social consequences like restrictions in driving and in pursuing certain professional careersSchulze-Bonhage A, Buller A: Unpredictability of seizures and the burden of epilepsy In: B. Schelter, J. Timmer, A. Schulze-Bonhage (Hrsg.): Seizure prediction in Epilepsy: From Basic Mechanisms to Clinical Applications Berlin: Wiley, 2008; 1-10.. Furthermore, the unforeseeable occurrence of seizures poses diagnostic problems and presently necessitates a continuous long-term treatment with anticonvulsants to prevent their occurrence. Seizure prediction could thus overcome one of the most disabling aspects of the disease by warning patients in time to avoid seizure-related risks. Asides it would open new avenues of specific, timely targeted and closed-loop treatments instead of long-term interventions mostly targeting interictal periods.Stacey WC, Litt B. Technology insight: neuroengineering and epilepsy-designing devices for seizure control. Nat Clin Pract Neurol. 2008; 4:190-201. Physiology of seizure generation Attempts towards seizure prediction assume that brain dynamics do not change abruptly but are preceded by gradual changes which can be extracted from analyses of appropriate physiological parameters . It has been advocated that such gradual transitions in brain dynamics can be expected only in epilepsies arising from circumscribed brain areas (“focal epilepsies”).Lopes da Silva F, Blanes W, Kalitzin SN, Parra J, Suffczynski P, Velis DN. Epilepsies as dynamical diseases of brain systems: basic models of the transition between normal and epileptic activity. Epilepsia 2003; 44: 72-83. Presently, the underlying pathophysiology of interictal transitions is unknown, as are parameters best suited for the detection of preictal changes. In principle, electrophysiological or chemical changes in brain activity, measures of brain metabolism and perfusion are considered as possible indicators. Most work has been concentrated on the field potential recordings of neuronal activity obtained at the level of scalp EEG, subdural or depth EEG. So far, truly prospective and statistically valid evidence that prediction of seizures is possible remains to be providedMormann F, Andrzejak RG, Elger CE, Lehnertz K. Seizure prediction: the long and winding road. Brain 2007; 130: 314-33., and there is an ongoing contest for its proveSchelter B, Feldwisch-Drentrup H, Timmer J, Gotman J, Schulze-Bonhage A: A common strategy and database to compare the performance of seizure prediction algorithms. Epilepsy Behav, 2010; 17: 154-156. Methodology of seizure prediction EEG-based seizure prediction uses linear or non-linear measures derived from the EEG. Based on a moving-window analysis, time-profiles of derived features are calculated. For predictions, thresholding of derived features is used; alternatively, seizure propensities can be derived from target ranges of features. Predictions have to be made for certain time periods (“seizure occurrence periods”, SOP) during which a seizure is assumed to occur. Furthermore, to distinguish predictions from detections, a minimum time interval between the derivation of a prediction and seizure onset has to be defined. The performance of seizure prediction algorithms is measured by its sensitivity to correctly predict seizures to occur during a SOP, the number of false prediction, and the statistical prove of superiority compared to random predictionsWinterhalder M, Maiwald T, Voss HU, Aschenbrenner-Scheibe R, Timmer J, Schulze-Bonhage A. The seizure prediction characteristic: a general framework to assess and compare seizure prediction methods. Epilepsy Behav 2003; 4: 318-25.Schelter B, Winterhalder M, Maiwald T, Brandt A, Schad A, Schulze-Bonhage A, et al. Testing statistical significance of multivariate time series analysis techniques for epileptic seizure prediction. Chaos 2006; 16: 013108.. Current status of seizure prediction Analyses performed in the 1990s focused on changes in the dynamics during time periods preceding seizures. Various methods, e.g.the largest Lyapunov exponentIasemidis LD, Sackellares JC, Zaveri HP, Williams WJ. Phase space topography and the Lyapunov exponent of electrocorticograms in partial seizures. Brain Topogr 1990; 2: 187-201, correlation densityMartinerie J, Adam C, Le Van Quyen M, Baulac M, Clemenceau S, Renault B, et al. Epileptic seizures can be anticipated by non-linear analysis. Nat Med 1998; 4: 1173-6., dynamical similarity indexLe Van Quyen M, Martinerie J, Baulac M, Varela F. Anticipating epileptic seizures in real time by a non-linear analysis of similarity between EEG recordings. Neuroreport 1999; 10: 2149-55.Le Van Quyen M, Martinerie J, Navarro V, Boon P, D'Havé M, Adam C, et al. Anticipation of epileptic seizures from standard EEG recordings. Lancet 2001; 357: 183-8., claimed to find characteristic preictal changes. When such changes were compared to fluctuations in interictal dynamics (as done on a first international database from 5 patients following the First International Seizure Prediction Workshop in Bonn, 2000Lehnertz K, Litt B. The First International Collaborative Workshop on Seizure Prediction: summary and data description. Clin Neurophysiol 2005; 116: 493-505.), most methods could not prove the statistical significance of such changesAschenbrenner-Scheibe R, Maiwald T, Winterhalder M, Voss HU, Timmer J, Schulze-Bonhage A. How well can epileptic seizures be predicted? An evaluation of a nonlinear method. Brain 2003; 126: 2616-26.,data mining using discrete finite automata have predicted seizures in rats by 6 seconds sR. Lewis and A. M. White. Deterministic Finite Automata in the Detection of Eeg Spikes and Seizures. in Proceedings of The Joint Venture of The Ninth International Symposium on Intelligent Data Analysis (IDA), Tucson, Arizona, pages 103–113, May, 9-21 2010.,Lai Y, Harrison MAF, Frei MG, Osorio I. Inability of Lyapunov exponents to predict epileptic seizures. Phys Rev Lett 2003; 91: 068102.Jouny CC, Franaszczuk PJ, Bergey GK. Signal complexity and synchrony of epileptic seizures: is there an identifiable preictal period? Clin Neurophysiol 2005; 116: 552-8.. Recent analyzes show that short time prediction (1 min) is possible with high accuracy if electrodes are located at the focal point where seizure develops Aur, D. Understanding the physical mechanism of transition to epileptic seizures, Journal of Neuroscience Methods Volume 200, Issue 1, 2011, Pages 80-85 . So far, evidence from retrospective analysis including limited interictal time periods suggest that some bivariate measures like phase synchronization are superior to random predictorsMormann F, Kreuz T, Andrzejak RG, David P, Lehnertz K, Elger CE. Epileptic seizures are preceded by a decrease in synchronization. Epilepsy Res 2003; 53: 173-85., but have insufficient sensitivity and specificity for clinical applications. Improved performances can result with consideration of circadian variability in EEG dynamicsSchelter B, Winterhalder M, Maiwald T, Brandt A, Schad A, Timmer J, Schulze-Bonhage A: Do false predictions of seizures depend on the state of vigilance? A report from two seizure-prediction methods and proposed remedies. Epilepsia, 2006; 47: 2058-2070, and by combining different measuresFeldwisch-Drentrup H, Schelter B, Jachan M, Nawrath J, Timmer J, Schulze-Bonhage A: Joining the benefits: Combining epileptic seizure prediction methods. Epilepsia, 2010 before print in the analysis of characteristic preictal alterations. The build-up of large databases containing recordings with sufficient duration and numbers of seizures is presently regarded as one important step to provide the growing number of research groups worldwide devoted to seizure prediction with adequate data for the development and analysis of new approaches for seizure prediction. Both the European Community (project “EPILEPSIAE”http://www.epilepsiae.eu) and the NINDS (USA) support such multicentric databases. See also * Emotiv Systems References *'Schelter B, Timmer J, Schulze-Bonhage A (eds.): Seizure Prediction in Epilepsy - From Basic Mechanisms to Clinical Applications, Berlin: Wiley - VCH, 2008' External links *Seizure prediction contest *Freiburg EEG Database with inter-ictal and pre-ictal recordings *European project EPILEPSIAE *4th International Seizure prediction workshop (Kansas City, USA) *3rd International Seizure prediction workshop (Freiburg, Germany) Category:Neurophysiology Category:Epilepsy